Research
Academic papers, new techniques, benchmarks, and theoretical findings in AI/LLM security.
Academic papers, new techniques, benchmarks, and theoretical findings in AI/LLM security.
44 items
The source is a bibliographic listing for a July 2026 paper in the Journal of Information Security and Applications, Volume 100, by Daniel Commey, Sena G. Hounsinou and Garth V. Crosby, titled "PUFZIN: Secure and scalable blockchain-IoT with PUFs and zero-knowledge proofs." The provided text contains no abstract, method or findings.
Researchers Abeer Qashou, Nurhidayah Bahar, and Hazura Mohamed published a paper in Computers & Security, available online 21 May 2026, on data security risks for mobile cloud computing in higher education. The source text provided contains only the title, publication details, and author list, so the study's method and findings are not described.
This article studies how a phased array radar should schedule its time across joint search and tracking tasks while under adversarial jamming. It formulates a multi-step dynamic optimization model with a receding-horizon strategy, solved by combining Pontryagin's maximum principle with ADMM. Numerical simulations reportedly show higher tracking accuracy, jamming resilience and total utility than several comparison methods.
Wang, Yan, Zhang and Wen propose DEGAN, a dual-enhanced GAN for detecting botnets in industrial IoT networks. The paper addresses imbalanced training data, and was published in the Journal of Information Security and Applications, Volume 100, in July 2026.
This paper studies leaky private information retrieval (L-PIR), where privacy leakage is measured by a pure differential privacy parameter called the leakage ratio exponent. The authors show that the active pure-DP constraints couple adjacent Hamming-weight layers of the random key, so the problem reduces to a layered optimization whose optimum is geometric across layers, requiring only cyclic permutations and assigning higher probabilities to lower-Hamming-weight keys. This yields an O(log K) leakage ratio exponent at fixed download cost D, improving on the previous Θ(K) exponent from prior work.
Vul-CTG is a multimodal framework for software vulnerability detection that combines code text and program graph representations. It builds enriched graphs from statement-level source code graphs and abstract code property graphs, aligns them with code text through a new architecture called CTG-Former, and uses contrastive learning and pre-training to resist noisy labels. On recent function-level datasets it reports an approximate 3% F1-score improvement over state-of-the-art methods.
Researchers propose TS-VulA, a three-stage vulnerability analysis framework for Industrial Internet of Things (IIoT) that combines attack graphs, ModernBERT-based SentenceBERT scoring of exploitation likelihood and criticality, and multi-layer heterogeneous network analysis of device importance. Experiments report average accuracy of 87.86% and average precision of 87.33% for the vulnerability assessment method, exceeding existing methods. Simulated case studies indicate TS-VulA outperforms prevailing vulnerability analysis methods.
The source is only publication metadata for a paper titled DWT-AMSA on robust image steganography, listing a July 2026 date, the Journal of Information Security and Applications (Volume 100), and five authors. It gives no question, method or findings.
Mahender Kumar, Ruby Rani, Gregory Epiphaniou and Carsten Maple published Adaptive Trust-Aware SOC Human–AI Teaming for resilient operations in Computers & Security, available online on 21 May 2026. The source text provided contains only publication metadata and no description of the method or findings.
The paper proposes FairRoP, a client selection scheme for federated learning that aims to improve fairness and robustness at the same time. It models the problem as a multi-objective optimization solved with an ε-greedy Thompson Sampling Multi-Armed Bandit, using fairness awareness, attack detection and q-Balance submodules. Experiments on real datasets reportedly show significant gains in fairness and robustness over state-of-the-art methods, and the approach integrates with other aggregation algorithms.
Researchers propose Trigger without Trace (TwT), a stealthy backdoor attack on text-to-image diffusion models. It uses syntactic structures as triggers to break semantic consistency and applies a Kernel Maximum Mean Discrepancy (KMMD) regularizer to align cross-attention distributions with benign samples. The method reports a 97.5% attack success rate and over 98% of backdoor samples bypassing three state-of-the-art detection mechanisms.
Researchers present Model Defense Variational Autoencoder (MDV), a method against model extraction attacks (MEA), where attackers build a locally trained clone of a deep learning model. MDV replaces realistic auxiliary data, which is often absent, unstable in effect, and leaves some categories less protected, with virtual auxiliary data. The authors say experiments show it addresses these three problems.